Covering Arrays to Support the Process of Feature Selection in the Random Forest Classifier

Vivas, Sebastián; Cobos, Carlos; Mendoza, Martha

doi:10.1007/978-3-030-13709-0_6

Cited by 3 publications

(2 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…CAs generally can be seen as a sampling mechanism in several contexts [56][57][58][59][60]. In this paper, we use CAs to sample solutions of multiple linear Diophantine equations with unit coefficients, where each row of the CA is used to construct a solution of an LDEU.…”

Section: Covering Arraysmentioning

confidence: 99%

Weight Vector Definition for MOEA/D-Based Algorithms Using Augmented Covering Arrays for Many-Objective Optimization

Cobos,

Ordoñez,

Torres-Jimenez

et al. 2024

Mathematics

Self Cite

View full text Add to dashboard Cite

Many-objective optimization problems are today ever more common. The decomposition-based approach stands out among the evolutionary algorithms used for their solution, with MOEA/D and its variations playing significant roles. MOEA/D variations seek to improve weight vector definition, improve the dynamic adjustment of weight vectors during the evolution process, improve the evolutionary operators, use alternative decomposition methods, and hybridize with other metaheuristics, among others. Although an essential topic for the success of MOEA/D depends on how well the weight vectors are defined when decomposing the problem, not as much research has been performed on this topic as on the others. This paper proposes using a new mathematical object called augmented covering arrays (ACAs) that enable a better sampling of interactions of M objectives using the least number of weight vectors based on an interaction level (strength), defined a priori by the user. The proposed method obtains better results, measured in inverted generational distance, using small to medium populations (up to 850 solutions) of 30 to 100 objectives over DTLZ and WFG problems against the traditional weight vector definition used by MOEA/D-DE and results obtained by NSGA-III. Other MOEA/D variations can include the proposed approach and thus improve their results.

show abstract

Section: Covering Arraysmentioning

confidence: 99%

Weight Vector Definition for MOEA/D-Based Algorithms Using Augmented Covering Arrays for Many-Objective Optimization

Cobos,

Ordoñez,

Torres-Jimenez

et al. 2024

Mathematics

Self Cite

View full text Add to dashboard Cite

show abstract

“…The tests with tweets showed the potential of the proposed method to reduce the indexes used in polarity detection. Vivas et al [45] based on a strategy that uses CAs, modified the process of selecting features for training trees in Random Forest (RF). They proposed using the rows of CAs as a feature selector instead of the random selector used by the random forest algorithm by default.…”

Section: Related Work a Wrappers For Feature Selectionmentioning

confidence: 99%

Wrapper for Building Classification Models Using Covering Arrays

et al. 2019

Self Cite

View full text Add to dashboard Cite

Wrapper methods are a type of feature selection method that finds a subset of variables to improve the performance of a classifier by removing redundant and irrelevant variables. The use of a wrapper implies that each time a candidate solution is explored, the classifier is evaluated on the quality measures selected (e.g. accuracy or precision). Though robust, this iteration across several candidate solutions can become computationally intensive and time-consuming. In this paper we propose a wrapper, that is based on binary Covering Arrays (CAs), and binary Incremental Covering Arrays (ICAs), that have been widely used for experimental design and fault detection in software and hardware testing. The new wrapper was evaluated with six classifiers on seven data sets. The results show that the CAs and ICAs with strength 6 significantly improve the performance and reduces the number of variables required by the classifier. A comparative analysis of the proposed method against wrappers based on other search approaches such as genetic algorithms (GA) and particle swarm optimization (PSO), shows that the proposed method yields results similar to GA, but not to PSO, with differences to PSO, in accuracy, which in the majority of cases is below 0.04. This lack of accuracy, by which the new wrapper fails to match PSO, is offset by the fact that the user does not need to fine tune algorithm parameters, such as velocity ranges, timing, cognitive coefficient, and social coefficient, while it is also much easier to program in parallel.INDEX TERMS Classification algorithms, covering arrays, random forest, support vector machines, genetic algorithms, particle swarm optimization. HUGO DORADO received the B.Sc. degree in statistics from the Universidad del Valle, Cali, Colombia, in 2013, and the M.Sc. degree in computing from the Universidad del Cauca, Popayán, Colombia, in 2019.He is currently a Research Assistant with the Research Group of Decision and Policy Analysis (DAPA), International Center for Tropical Agriculture (CIAT). His role has been focused in analyzing observational information in agriculture by using analytical methodologies for detecting patterns associated with either high or low yields. His research interests include data mining, computational intelligence, and statistical science.

show abstract

Amazonian cacao-clone nibs discrimination using NIR spectroscopy coupled to naïve Bayes classifier and a new waveband selection approach

Castro¹,

De-la-Torre

Ávila-George

et al. 2022

Spectrochimica Acta Part A: Molecular and Biomolecular Spectros

View full text Add to dashboard Cite

Covering Arrays to Support the Process of Feature Selection in the Random Forest Classifier

Cited by 3 publications

References 12 publications

Weight Vector Definition for MOEA/D-Based Algorithms Using Augmented Covering Arrays for Many-Objective Optimization

Weight Vector Definition for MOEA/D-Based Algorithms Using Augmented Covering Arrays for Many-Objective Optimization

Wrapper for Building Classification Models Using Covering Arrays

Amazonian cacao-clone nibs discrimination using NIR spectroscopy coupled to naïve Bayes classifier and a new waveband selection approach

Contact Info

Product

Resources

About